introduction - ericgriggs, olson, beasley, & gorman, 1995). the dunn and dunn learning-style...

A Meta-analysis of Dunn….

A Meta-analysis of Dunn and Dunn Model Correlational

Research with Adult Populations

by

Christine Mangino

____________________________________________ CHRISTINE MANGINO is an assistant professor in the Early Childhood Department at

Hostos Community College, Bronx, New York and an adjunct professor in the

Instructional Leadership program and Childhood Education Department at St. John’s

University, Jamaica, New York. 1583 Chapin Avenue, Merrick, NY 11566 (516) 546-

1996 [email protected]

A Meta-analysis of Dunn…. 2

Abstract

The purpose of this investigation was to conduct a quantitative synthesis of

correlational research that focused on the Dunn and Dunn Learning-Style Model and was

concerned with adult populations. A total of 8,661 participants from the 47 original

investigations provided 386 individual effect sizes for this meta-analysis.

The mean effect size was r = .258. This suggested that the learning-style elements

had a medium effect on the 30 independent variables explored. The largest effects were

on Discipline (r = .363), Achievement (r = .351), Decision-Making Groups (r = .343),

Age (r = .326), Ethnicity (r = .311), Right Brain/Left Brain (r = .303), and Gender (r =

.288).

Six of the 15 independent variables were discovered to have moderated the

results. These included the assignment to research sample, demographic region,

publication type, school level, type of statistics utilized to calculate effect size, and

university.

Publication bias was not revealed through correlations of sample sizes and effect

sizes. Calculation of a Fail Safe N statistic determined between 1,439 and 1,644 studies

supporting the null hypothesis would be necessary to reverse the conclusion that

individual’s preferred learning-styles were significantly related to the categorical

variables examined.

Keywords: learning styles, meta-analysis, higher education, Dunn and Dunn Model,

adults


A Meta-Analysis of Dunn and Dunn Model Correlational

Research with Adult Populations

Most people have heard the statistic that seven babies are born every minute in the

United States, but few are aware of the astonishing detail that one journal article is

published every 30 seconds in the sciences (Mahoney, 1985). That fact makes it

impossible for individuals to find, read, and analyze all possible publications on a single

topic of interest. To further compound the problem, the research concerning the same

topic is usually diverse and complex, and may appear in a variety of journals with varied

foci. For instance, while a researcher at the University of Mississippi examined the

characteristics of freshman students of Alcorn State University, another researcher at

Boston University investigated adult learners at an Alaskan oil industry corporation, and

yet another researcher at St. John’s University in New York examined non-traditional

college students in multiple treatments. These researchers all used the Dunn and Dunn

Learning-Style Model (1972, 1992, 1993, 1999) as the cornerstone of their investigation,

but each was examining different age, achievement, and geographical populations and

very different variables.

A simple search of learning style on Educational Resources Information Center

(ERIC) revealed the existence of 4,019 articles. Thus, there appears to be a need to

quantitatively synthesize this voluminous research to integrate its findings, analyze

relevant theories, resolve conflicts that appear in the literature, and identify central issues

and findings for future investigations (Cooper & Hedges, 1994, p. 5).


Overview of the Dunn and Dunn Learning-Style Model

During the late 1960s, researchers began to examine the alternative ways in which

individuals learn. Their discoveries evolved into what is known today as learning style

(Tendy & Geiser, 1998-9). Learning style is comprised of biological and developmental

characteristics that make the identical instructional environments, methods, and resources

effective for some learners and ineffective for others (Dunn & Dunn, 1972, 1992, 1993;

Thies, 1979, 1999/2000). Compared with other learning-style approaches, the Dunn and

Dunn Learning-Style Model: “(a) includes greater comprehensiveness, (b) is more

extensively researched, and (c) demonstrates higher levels of consistent effectiveness”

(Given, 1997-8).

According to Dunn and Dunn (1993), learning style is the way students begin to

concentrate on, process, internalize, and remember new and difficult academic

information. Because at every age, people learn more, do so more easily, and retain it

better when they use their learning style, their styles are actually their strengths (Dunn,

Griggs, Olson, Beasley, & Gorman, 1995).

The Dunn and Dunn Learning-Style Model is based on the theory that:

1. most individuals can learn;

2. different instructional environments, resources, and approaches respond to

different learning-style strengths;

3. everyone has strengths, but different people have very different strengths;

4. individual instructional preferences exist and can be measured reliably (Burke,

Guastello, Dunn, Griggs, Beasley, Gemake, Sinatra, & Lewthwaite, 1999-2000);


5. given responsive environments, resources, and approaches, students attain

statistically higher achievement- and attitude-test scores in congruent, rather than

incongruent treatments (Dunn & Dunn, 1992, 1993; Dunn & Griggs, 2003).

According to this model, learning style is divided into five strands called stimuli. The

first stimulus strand consists of biologically-imposed environmental elements (Thies,

1979, 1999-2000). These include preferences for Sound versus Quiet, low versus bright

Light, warm versus cool Temperatures, and Formal versus Informal Seating. The

combination of Light and seating Design affect approximately 70 percent of people

(Dunn & Dunn, 1998, p. 8). Although room Temperature and Sound affect only a small

percentage of learners, for those who have a strong preference for either cool or warmth

or quiet versus sound while concentrating, these elements become critical for functioning

effectively. Such people become distracted by this need and are unable to concentrate if

the temperature or acoustics in the room do not match their biological preferences (Dunn,

Thies, & Honigsfeld, 2001).

The model’s second stimulus strand includes the emotional elements of

Motivation, Persistence, Responsibility, and Structure. Whereas the element of

persistence is biologically imposed, the others are developmental (Thies, 1979, 1999-

2000). Persistence is the need to complete a task before taking a break versus the need to

take many breathers while working on a task. Motivation is concerned with whether or

not a person is internally versus externally motivated, whereas Responsibility is denoted

by whether a person is conforming or nonconforming, and structure refers to individuals’

needs for internal versus external direction.


The sociological elements of Learning Alone, in a Pair, with Peers, as part of a

Team, or with an Adult who is either authoritative or collegial, and the need to work in a

Variety of Ways versus in a routine, comprise the third stimulus. These elements are

developmental (Thies, 1979, 1999-2000) meaning that they tend to change over time in

predictable patterns.

The physiological strand is composed of the elements of Perceptual Preferences,

Intake, Time, and Mobility. The four modalities of perceptual strengths are: auditory

(remembering ¾ of what is heard): visual (remembering ¾ of what is read or seen):

tactual (remembering ¾ of what is written or manipulated with the hands); and

kinesthetic (remembering ¾ of what is experienced). The idea that individuals remember

differently the complex information they learn by hearing, reading, seeing, tactually

manipulating, or experiencing may be one of the major findings of this era (Dunn &

Dunn, 1998, p. 10). Perceptual Strength and Time of Day each affect approximately 70

percent of all people. Intake is the need to snack while learning and Mobility is the need

to be pacing, rocking, or changing positions at frequent intervals while learning.

The fifth stimulus strand incorporates the psychological elements. These include

Global versus Analytic processing, Hemisphericity, and Impulsive versus Reflective

behaviors. The elements of Hemisphericity and Global/Analytic essentially appear to

parallel each other (Dunn, Beaudry, & Klavas, 1989). Both refer to a preference for

simultaneous versus sequential mental processing.

No one is affected by all 21 elements. (See Figure 1.) Most people are impacted

by somewhere between 5 and 14, although some individuals are affected by as many as

16 (or more); many are impacted by fewer (two to six). All elements are important and


contribute differentially to how well each adult concentrates on, processes, internalizes,

and retains new and difficult information (Dunn & Dunn, 1998, p. 8).

It is difficult to identify learning style accurately without a reliable instrument

(Beaty, 1986; Dunn, Dunn, & Price, 1977; Marcus, 1977). Several reliable and valid

assessments that identify individual learning-style preferences have been developed.

These inventories are self-reporting questionnaires that identify individual preferences.

The Learning-Style Inventory (LSI) (Dunn, Dunn, & Price, 1975, 1978, 1984, 1986,

1987, 1989, 1990, 1996) is based on the Dunn and Dunn Learning-Style Model and

different age-appropriate versions analyze the conditions under which students in grades

3 through 12 prefer to learn. The LSI also is appropriate for academically underachieving

college freshmen. The Productivity Environmental Preference Survey (PEPS) (Dunn,

Dunn, & Price, 1981, 1986, 1989, 1990, 1991, 1993, 1996) is appropriate for adult

students and Building Excellence (BE) (Rundle & Dunn, 1996-1999) is a globally

formatted instrument for corporate personnel. Once learning styles have been identified,

“instructors can estimate the approach(es), method(s), and sequence(s) that are likely to

make learning relatively comfortable for each person” (Dunn & Griggs, 2000, p. 19).

Purpose of the Study

There have been more than 800 studies conducted at more than 120 institutions of

higher education with the Dunn and Dunn Model (Research on the Dunn & Dunn Model,

2004). The findings from this research were too numerous and diverse to be fully

understood through a narrative review of the literature. Instead, a quantitative synthesis of

this body of research led to a better understanding of the overall impact of learning-style

characteristics.


One way to translate results from varied studies to a common metric and to

statistically explore relationships between investigated characteristics and findings is

through meta-analysis. A meta-analysis refers to an analysis of analyses or to the

statistical analysis of a large collection of results from individual investigations for the

purpose of integrating the findings (Glass, 1976). This strategy provides a viable process

for comparing the differences among studies in treatments, settings, measurement

instruments, and research methods that make their findings difficult to compare. Even

frequent replications can prove inconclusive if these variables are interchanged.

Literature on a topic may be so extensive as to obscure trends with an overwhelming

amount of information (Bangert-Downs & Robert, 1991).

Therefore the purpose of this research was to conduct a quantitative synthesis of

correlational studies elicited from educational journals and doctoral dissertations in which

the Dunn and Dunn Learning-Style Model served as the cornerstone between 1980-2003

and which used post-high school populations.

Faith, Allison, and Gorman (1997) described the three basic goals of meta-

analysis. The first goal is to find the average effect size of all studies to determine the

overall effectiveness of the construct under investigation. The second goal is to determine

the homogeneity or heterogeneity of that average. If the individual effect sizes are close

to the average effect size, “then there is little reason to suspect that other variables are

related to variations in effect size” (p. 272). If there is considerable variability around the

average, usually more than 25% of the population effect size, then the researcher searches

for “variables that moderate effect sizes” (p.246). Thus, a third goal of meta-analysis is to

identify the variables that lead to larger or smaller effects.


Method

Data Collection A comprehensive and methodical literature search was conducted to locate both

published and unpublished correlational investigations using post-high school

populations between 1980-2003 and based on the Dunn and Dunn Learning-Style Model.

Studies were located through several approaches.

The Learning-Style Network offers two compilations of research on this model.

Research on the Dunn and Dunn Model (2003) itemizes books, articles, and dissertations

focused on the Dunn and Dunn Model. Annotated Bibliography of Research (2003)

contains summaries of books, articles, and dissertations on several models of learning

style. Relevant research for this meta-analysis was found using these comprehensive

sources.

Next, electronic databases were systematically scanned for both published and

unpublished research including the terms learning style and Dunn. This computer-based

search identified relevant studies contained in Educational Resources Information Center

(ERIC), Proquest, Cumulative Index to Nursing and Allied Health Literature (CINAHL),

Education Full Text, Government Printing Office (GPO) on SilverPlatter, Library and

Information Science Abstracts, Library Literature and Information Science, and

PsychINFO.

After computer databases were searched, it was necessary to conduct an “ancestry

analysis for older literature by checking reference lists of retrieved publications” (Faith,

Allison, & Gorman, 1997, p. 248). This process also has been titled footnote chasing

(Cooper & Hedges, 1994, p. 46-47). Reference lists of research papers were searched to


locate relevant studies. The publisher of Building Excellence was contacted to gather

research conducted with this instrument.

For a study to be included in this meta-analysis, it had to have met specific

inclusion criteria. This review and subsequent investigation focused on the 47

correlational studies that met all the criteria.

Initially, to be included, the research must have been a correlational study.

Experimental, quasi-experimental, theoretical, and descriptive studies were excluded.

Second, the investigation must have utilized one of three instruments based on the

Dunn and Dunn Learning-Style Model. The sample also must have included college

students or adults.

Finally, the study must have reported enough statistical information to estimate

effect sizes. To calculate effect size, the investigation must have reported the number of

participants and summary statistics, such as F ratios, means and standard deviations, t-

tests, or significance levels.

Coding Study Characteristics

Two different types of variables were considered when coding studies for

this meta-analysis. The dependent variables were effect size values. The independent

variables were study and design characteristics that may have influenced the magnitude

of these effect sizes (Lipsey & Wilson, 2001).

Once the coding forms were developed, several studies were coded to be certain

all characteristics were included and the codebook was explicit enough to enhance

reliability. Next, a statistician familiar with meta-analyses was consulted for his expertise

on the appropriateness and details of the coding forms. Once the forms were determined


effective, this researcher coded the studies following the codebook. To increase

reliability, two additional coders randomly chose 10 % of the studies to code. Intercoder

reliability then was calculated at 93% agreement.

Statistical Analysis

The first stage for this part of the analysis is the combination of study results.

Several statistical packages were used to analyze the data.

Prior to combining studies, it is recommended to assign more weight to studies

with larger sample sizes (Schmidt & Hunter, 1977; Hunter et al., 1982; Hunter &

Schmidt, 1990). The weighted and unweighted average effect sizes, with corresponding

confidence intervals, were calculated.

Several methods were used to interpret mean effect-size values and to evaluate the

significance of the average effect-size values. The procedures included computing 95%

confidence intervals and using Cohen’s (1977, 1988, 1992) definitions for small,

medium, and large effect sizes for r and d calculations.

Confidence intervals indicate the range within which the population mean is

likely to be, given the observed data. Wolf (1986) suggested a 95% confidence interval

be calculated to interpret an effect size and to examine whether it includes zero. “It would

be desirable for the average effect size not to encompass zero, to be positive there is a

significant effect across these studies,” (p.27).

These procedures fulfilled the first goal of meta-analysis, which is the combining

of effect sizes. The various measures of average effect sizes and their interpretations

revealed the correlations among learning-style elements and the categorical variables

examined. The following procedures determined the heterogeneity of the effect sizes and


examined whether the selected study characteristics and design features moderated the

correlations of learning-style preferences and the variables investigated. These fulfilled

the second and third goals of meta-analysis, the determination of homogeneity and the

search for moderating variables.

Moderating Variables

Once the various effect sizes are averaged into a mean value, it is important to

determine if they all estimate the same population effect size (Hedges, 1982; Rosenthal &

Rubin, 1982). The homogeneity test assesses the null hypothesis that the between-studies

variance in effects is no greater than would be expected due to sampling error alone, or

25% of the population mean.

Three indicators of homogeneity were calculated and examined. All three tests

rejected homogeneity. Mean effect sizes for learning-style elements and specific

variables investigated were varied enough to be described as heterogeneous.

Consequently, it was essential to search for variables that moderated effect sizes.

Publication Bias

According to Glass, McGaw, & Smith (1981), a major criticism of meta-analyses

is that they are dependent on the findings that researchers report. A meta-analysis’

“findings will be biased, if as surely true, there are systematic differences among the

results of research that appear in journals versus books versus theses versus unpublished

papers” (p. 226). In general, statistically significant findings are more likely to be

published than not. As this was an anticipated issue, publication type was considered a

moderating variable in the coding system for this meta-analysis. Primary studies were

coded to differentiate among dissertations or theses, published journal articles, both


dissertations and published article, and conference papers. Potential publication bias was

assessed in three ways.

Light and Pillemer (1984) introduced the funnel plot for the graphic detection of

publication bias. A funnel plot is a scatterplot of sample size versus estimated effect size

for a group of studies. Since small studies typically will show more variability among the

effect sizes than will larger studies, and there will be typically fewer of the latter, the plot

should look like a funnel. “When publication bias is present, the funnel will look as if it

has been cut off at some point or a variation in the density of points will be observed,”

(Cooper & Hedges, 1994, p. 393). This procedure was conducted and did not suggest

bias. Correlations between sample sizes and effect sizes were examined during the

cluster analysis, but since low correlation was revealed, publication bias was not

probable.

The third assessment was the calculation of the Fail Safe N through Schwarzer’s

Meta-analysis Program (1986, 1996). This procedure, originally developed by Rosenthal

(1979), estimates the number of unpublished studies reporting null results needed to

reduce the cumulated effect across studies to the point of nonsignificance (Lipsey &

Wilson, 2001. p. 166). High values for this statistic indicated strong results and a reduced

probability that unpublished results could change the overall conclusion.

Results

Mean Effect Sizes

Fifty-nine studies were identified that were based on the Dunn and Dunn model,

were correlational, and focused on adult populations. Twelve of those investigations did

not provide appropriate statistical information to calculate reliable effect sizes. The


remaining 47 explorations were accepted and included in this research. The total sample

size was 8,661 and the total number of effect sizes was 386.

The POWPAL program (Gorman, Primavera, & Karras, 1995) was used to

convert individual findings into effect sizes r. Selected investigations provided multiple

effect sizes within one study. The range included five studies that produced a single

effect size to one investigation that provided 38 effect sizes. The effect sizes ranged from

.009 to .785. The mean effect size was .258. Following Cohen’s definitions for small,

medium, and large effect sizes (ES), this meta-analysis included 204 small ES, 140

medium ES, and 42 large ES. Those effect sizes that were .249 and less were categorized

as small; effect sizes that were within .250 - .399 were labeled as medium; and effect

sizes equal to or above .400 were considered large.

The largest value for r was the Fisher's Z transformation of r = .269 which was

weighted by sample size and the smallest value for r was the Schmidt-Hunter weighted r

of r = .236. All the confidence intervals were far from zero (Table 1).

(INSERT TABLE 1 ABOUT HERE.)

This supported the assumption that there were significant relationships between

individuals’ learning styles and the variables examined in this investigation, such as

academic level, achievement, age, attitude, computer usage, discipline, ethnicity, gender,

grade-point average (GPA), and profession. Although these were the variables from the

original research questions, several other variables were revealed once the studies were

gathered and examined closely (Table 2).

Achievement was measured by the researchers through the results on standardized

tests, including a reading test and a nursing examine for certification. Higher-achieving


students revealed preferences for learning in Several Ways (r = .312), an Authority-figure

present (r = .281), the need for Structure (r = .382), no Sound (r = .400), a Formal Design

(r = .470), and tended to be more Motivated (r = .249) then lower-achieving students who

revealed a strong preference for Light (r = .450). Examining the perceptual strengths,

high achievers had a large effect for Kinesthetic (r = .637) and medium effects for

Auditory (r = .395) and Tactual (r = .339), whereas the low achievers had a small

preference for Tactual (r = .228) and a non-preference for Visual (r = .419) (Table 3).

Age was the most examined variable by researchers. Subjects under 30 years of

age tended to be more Kinesthetic, prefer to learn with Peers or an Authority-figure in the

Afternoon, be Parent-Motivated, and in need of Structure. Participants over 30 years of

age preferred to Learn Alone, were self Motivated and teacher motivated, preferred to

learn in the Morning with Structure and Light, and were Responsible, Visual, and

Tactual. Individuals over 55 years of age were Self-Motivated, Responsible, and Tactual,

preferred learning with Peers in the Late Morning or with an Authority-Figure present,

and had a large effect size for Structure (r = .785) (Table 4).

Attitude was measured separately with job/school satisfaction and with attitude

toward studying. With respect to job or school satisfaction, individuals who were

satisfied revealed a small mean effect for the need to Learn Alone with no Sound.

Subjects who were not satisfied disclosed small effects for Persistence, Responsibility,

and Motivation. With regard to study attitude, students who had a positive attitude were

Responsible, Motivated, Reflective, Visual, and preferred to Learn Alone. Those with a

negative attitude appeared to prefer Intake and Mobility while learning (Table 5).


Computer usage resulted in seven individual effect sizes with a mean effect of r =

.216 for individuals with high computer usage. Most of the elements were associated with

Perceptual Strengths, although the largest effect was evidenced for Morning learners,

with a medium mean effect size of r = .291 (Table 6).

The discipline, or academic major in which students were enrolled, evidenced the

largest mean effect size. Education majors revealed preferences for three of the four

perceptual modalities, Kinesthetic (r = .310), Visual (r = .332), and Tactual (r = .313).

They also tended to be Responsible and needed Mobility. Nursing majors were

Kinesthetic (r = .388) and Visual (r = .420). They also preferred the presence of an

Authority figure, working with Peers in the Morning and Afternoon, Intake and Light,

and were Motivated and Reflective. Music majors had large effect sizes for Auditory (r =

.524) and Kinesthetic (r = .612) instructional methods. They also preferred the presence

of an Authority figure, working with Peers in the Morning, Intake, and Mobility while

learning new and difficult information. Business majors revealed a small effect size when

correlated with all the elements but none of the researchers provided appropriate statistics

to calculate effect sizes for individual elements (Table 7).

Ethnicity had an overall mean effect size (r = .311) which was one of the larger

effects from the variables examined. Each of the researchers examined different groups

and subgroups accounting for six different groupings in this meta-analysis. Caucasians

revealed medium size effects for Design, Morning, Auditory, and Visual. Hispanics

unveiled medium effect sizes for Intake, Auditory and Kinesthetic. Non-Caucasians

disclosed a small effect for Afternoon learning (Table 8).


Gender had a medium size effect when it was correlated with all the elements (r =

.358) and revealed a correlation with most of the elements. The data suggested that males

preferred Tactual and Auditory instructional activities, learning in the Afternoon with

Peers, Mobility, and learning in Several Ways. Adult males also tended to be Persistent

and Responsible and males disclosed a large effect size for Design. Females revealed a

preference for Structure, an Authority-figure present, Kinesthetic and Visual modalities,

Temperature, Design, learning in Several Ways, and Mobility. Females tended to be

Persistent, Motivated, Responsible, Morning learners, and preferred Learning Alone.

Females unveiled a large effect size for Intake (Table 9).

Grade-point average (GPA) evidenced an overall mean effect size of r = .191.

When examined closer, one medium and eleven small mean effects emerged. When GPA

results were combined with the data from achievement and entrance exam scores, the

mean effect size was r = .275. The table outlines the mean effect sizes for GPA by itself

and when it was combined with the other two variables. This combination, which

together can be considered achievement, shared many of the same elements. This

similarity, helped support the results that achievement is correlated with learning-style

elements. Students with a higher GPA preferred Kinesthetic activities, Structure,

Learning Alone, Intake, no Authority-figure present, and learning in Several Ways. They

were Motivated and Responsible and did not prefer Tactual instructional methods.

Students with lower GPAs preferred learning in the Afternoon and with Sound present

while learning (r = .127). When GPA was combined with achievement on standardized

tests and entrance exam scores, students with high achievement revealed a large effect for

Kinesthetic learning (r = .472). They also appeared to be Responsible, Motivated,


needing Structure, Intake, and to learn in Several Ways. They tend to be Auditory but not

Visual. They prefer both working Alone and with Peers. Students with low achievement

appear to be Visual and prefer learning in the Afternoon (Table 10).

Numerous researchers had examined the relationship between the profession

subjects chose and learning styles. Educators revealed small effect sizes for Auditory and

Visual learning, Design, Intake, Mobility, an Authority-figure present, learning in Several

Ways, and were Responsible and Persistent. Corporate employees disclosed small effects

for each of the four perceptual modalities, preferred learning in Several Ways with Sound

and Intake. They also preferenced Learning Alone, with an Authority-figure present, and

in the Afternoon. Nurses tended to be Kinesthetic learners who needed bright Light, and

to learn with Peers. The correlation between nurses and peers was the only medium sized

effect for this variable. Paralegals revealed a non-preference for Tactual learning methods

(Table 11).

(INSERT TABLES 2-11 ABOUT HERE.)

Moderating Variables

Calculations involving the mean effect sizes fulfill the first part of a meta-

analysis. A major criticism of this statistical procedure is of those researchers who

conclude their investigation at this point. The most valuable part of a meta-analysis is the

search for moderating variables. Therefore, that was the second part of this analysis.

Most of the study and design characteristics indicated significance at the p <. 000 levels

when post hoc tests were conducted for homogeneity. This first indicator rejected

homogeneity and suggested heterogeneity.


Observed effect-size variances were calculated and decomposed into two

components of explained sampling error variances and unexplained population or

residual variances. A second indicator of homogeneity suggested that sampling error

variance should comprise at least 75 % of the total variance. When Schmidt-Hunter was

utilized, the percentage of observed variance accounted for by sampling error was

43.06%. Fisher’s Z Transformation calculated the sampling error variance as 42.36%.

Both of these calculations were less then the needed 75%; therefore, homogeneity was

again rejected and heterogeneity was evidenced.

A third indicator of homogeneity was that residual or population variance should

not exceed 25% of the population effect size. In all procedures, the residual standard

deviation was more than one-quarter of the population effect size. The residual standard

deviation values ranged from .079 to .08, whereas the suggested limit of 25% of the

population effect sizes ranged from .059 to .06. This third test, yet again, rejected

homogeneity and corroborated heterogeneity.

Given that all three indicators rejected homogeneity for effect sizes, the mean

effect sizes were sufficiently varied to warrant description as being heterogeneous.

Hence, it was necessary to search for variables that moderated these effect sizes.

The mean effect sizes for the type of data used to calculate effect size were

significantly different among the four categories (F = 6.886; p = .000). The mean effect

size was found to be greater when means and standard deviations were utilized to

calculate effect size (r = .329). Tukey’s Multiple Comparisons revealed significant

differences between means and standard deviations and t-tests and F-ratios (p < .000);

means and standard deviations and significant values (p < .001); and means and standard


deviations and chi-squares approached significance (p < .055). Hence, the type of data

used to calculate effect size appeared to have been a moderating variable in this

investigation.

Publication Type had significantly different mean effect sizes among the four

categories using the regular ANOVA (F = 18.495; p = .000) and Kruskal-Wallis ANOVA

(p = .000). The effect sizes derived from research that reflected both a dissertation and a

journal article (r = .413) had significantly different mean effect sizes from those that were

strictly dissertations (r = .258), journal articles (r = .224), or conference-presentation

papers (r = .173) at the p < .000 level for each category. There was no significant

difference between the mean effect sizes revealed in dissertations and journal articles (p =

.072) or between effect sizes disclosed in a conference paper and a dissertation (p = .156)

or a journal article (p = .622). Research that emerged originally from a dissertation and

was later published as a journal article had significantly higher correlations between

learning-style elements and specific variables, compared with other forms of research.

The large variety of University Affiliations was divided into regions rather than

specific universities. Significant differences among the mean effect sizes were revealed

(F = 2.901; p = .014). Closer examination disclosed a significant difference between

mean effect sizes derived from research produced at universities in the Northeast (r =

.290) and the mean effect sizes evolving from research produced at universities in the

Southeast (r = .244) at the p = .011. There were no other significant differences among

the other regions.

School level revealed significant differences between the mean effect sizes (F =

3.373; = .005). Further examination unveiled significant differences between the mean


effect size for freshmen (r = .228) and undergraduates (r = .280) at the p = .013 level, and

the differences between freshmen (r = .228) and adults (r = .273) approached significance

at the p = .058 level.

The procedures for selecting sample populations for the research included in this

study also revealed significance in the ANOVA (F= .523; p = .000). Volunteer samples

had the largest mean effect sizes (r = .305) compared with other categories, Random

assignment (r = .238), using an Entire Sample Population (r = .239), a Stratified Sample

(r = .249), and when the sample assignment was Not Specified (r = .251). Volunteer

mean effect sizes were significantly different from each of these groups at the p = .007,

.000, .018, .010, respectively.

The variables of assignment, publication type, school level, type of data utilized

to calculate effect size, and university region were moderating variables in this

investigation. Consequently, research that utilized means and standard deviations, began

as a dissertation and later became a published journal article, was conducted in a

northeastern region, and was concerned with volunteer undergraduates, revealed the

highest mean effect sizes.

Three sources of evidence evaluated publication bias for mean effect sizes. First, a

scatterplot compared effect size and sample size. The scatterplot provided evidence that

publication bias was not likely in this meta-analysis. Second, the correlations between

sample size and effect size were examined during the cluster analysis. It was shown that

one of the largest mean effect sizes (r = .491) had a relatively small sample size (N = 46)

and, conversely, one of the smallest mean effect sizes (r = .109) emerged from one of the

largest sample sizes (N = 144). This, too, provided evidence that publication bias was


unlikely. A third procedure, the calculation of Fail Safe N values estimated that between

1,439 and 1,644 studies supporting the null hypothesis would be necessary to reverse the

conclusion that individuals’ preferred learning styles were significantly related to specific

variables.

A total of 8, 661 adult participants from 47 correlational investigations provided

386 individual effect sizes for this meta-analysis. Although five moderating variables

influenced the outcome, the results of this investigation supported the hypothesis that

learning styles are significantly correlated with specific attribute variables.

Discussion

Research has repeatedly demonstrated the impact of accommodating individual

learning styles on achievement and attitudes (Dunn & Griggs, 2003). More recently,

research also revealed their impact on behavior (Fine, 2002; Oberer, 1999, 2003). This

investigation has demonstrated the impact learning style has on other aspects of an

individual’s life. Learning style is related to the discipline a student chooses, the

profession an adult enters, the school and program an individual decides to attend, a

person’s satisfaction with school, teacher or work, the amount of time an individual uses

a computer, the degree a student attains, and a person’s study habits and attitudes.

Conversely, a person’s genetic composition also affects her learning style. An

individual’s gender, ethnicity, age, field dependence, intuition, feeling, decision-making

abilities, learning disability, and whether or not she is shy or extroverted is related to an

individual’s learning style.

Although it would be easy to state that all students with good study habits are

motivated, especially with its large effect size (r = .707), it is still a generalization. It is


important to understand the 386 individual effect sizes are tendencies in learning styles

and the variables examined. The tendencies remind us that each student, employee, or

friend is an individual whose needs and preferences should be accommodated for optimal

success. Many of the correlations between learning-style elements and specific variables

had a single effect size encompassing the mean effect size, but it was still a significant

correlation. Also, each variable was significantly correlated with numerous learning-style

elements. It is key to have an understanding of learning styles and to have an

understanding of how an individual’s learning style has at least 386 different effects on a

person’s being. Twenty-nine variables were investigated by researchers, although it

probably does not end with those twenty-nine variables. Most likely, if each of these were

significant, there are more variables that have yet to be investigated.

Teachers and employers need to test adults with whom they interact for learning

styles. The Learning-Styles Model is multidimensional with various stimuli and many

elements; an instrument is needed to diagnose an individual’s learning style. Selected

learning-style instruments have proven to be valid and reliable. The results of this

investigation support that fact. There was virtually no difference between the mean effect

sizes by the instrument utilized. Studies that tested their populations with the Learning

Style Inventory had a mean effect size of r = .2574, whereas studies that utilized the

Productivity Environmental Preference Survey revealed a mean effect size of r = .2580,

for a difference of r = .0006.

An individual’s learning-style profile should be used for all aspects of college

living and employment. The discipline in which a student was enrolled had the largest

effect size when correlated with specific learning-style elements (r =.363, p = .000).


Overall achievement also was highly correlated with learning-style traits (r =

.351, p = .000). Professors should be required to accommodate their students’ learning

styles to ensure academic success. Society would agree to professors’ lack of

professionalism if they admitted to teaching only to those students with curly hair. It

seems illogical to permit professors to only teach to one type of learner. We need to find

methods to help those students, with learning-style traits that were not correlated with

high achievement become successful learners. For instance, low achievers were found to

be tactual rather than visual learners. It does not seem fair that because they learn

tactually but teachers teach by talking, that this group is prevented from becoming high

achievers. Educators need to change that.

We use different methods for growing cabbages and

azaleas. And there is no problem over which is better; one

isn't right and the other wrong. Anyone would call a farmer a

fool who planted them in the same place and gave them the

same fertilizer, (amount of) sun, and water. We value each,

and knowing that they will not thrive unless (their) needs are

met, we respect their different natures and accept their special

requirements. When we respect the differences we know exist

in people, and when we value the contributions to be gained

by those differences, we shall ...provide for their nurture and

cultivation.... (Reckinger, 1979).

The lack of publication bias is also a strength of this model. Although more than

800 studies support the research on the Dunn and Dunn Model, a majority of the research


is in the form of dissertations. The results of Hypothesis Three strongly support the

statement that there is no statistical difference in the research found in doctoral

dissertations and research published in journal articles.

It was revealed that research that began as a dissertation and was later published

as a journal article had a larger mean effect size when compared with research that was

only in the form of a dissertation or only in the form of a journal article. It may be

possible that the requirement of being screened by two separate filter systems,

dissertation committees and editors of journals, produces higher quality research.

The quality of the research was also an issue in this meta-analysis. Eleven studies

were excluded from the analyses for failing to supply adequate statistical information. A

few studies failed to acknowledge how the sample population has been gathered. As was

shown in the analyses, the type of sample was important when interpreting the results.

Therefore, it is important for researchers to explain how their samples are derived.

Medical research has developed a database, the Cochrane Collaborative, which is

responsible for reviewing all medically related research and incorporating it into one

system. It is an official means of uniformly reporting evidence-based research results.

The goal is to have everyone involved able to find relevant research, to allow researchers

to build on one another’ work rather than continually repeat studies because they are

unaware of previous work, and to have all results accessible for medical use. It may be

helpful to adopt this type of system to education.


References

*Adams, C. B. (2003). An analysis of the relationship between learning styles and the

growth in technology abilities among teacher education majors. (Doctoral

dissertation, University of Mississippi). Dissertation Abstracts International,

64(05), 1509A.

Annotated Bibliography of Research (1992, 1995, 1999, 2000, 2002). Jamaica, NY: St.

John’s University Center for the Study of Learning and Teaching Styles.

*Bailey, G.K. (1988). Examination of the relationship between hemispheric preferences

and environmental characteristics of learning style in college students. (Doctoral

dissertation, The University of Southern Mississippi). Dissertation Abstracts

International, 49(08), 2151A.

Bangert-Drowns, Robert L. & Rudner, Lawrence M. (1991). Meta-analysis in educational

research. Practical Assessment, Research & Evaluation, 2(8).

Beaty, S.A. (1986). The effect of inservice training on the ability of teachers to observe learning styles of students. (Doctoral dissertation, Oregon State University). Dissertation Abstracts International, 47, 1998A. *Bovell, C. (2000). Analysis of learning styles of older students enrolled in non-

traditional college programs. (Doctoral dissertation, St. John’s University). Dissertation Abstracts International, 62(11), 3656A.

Burke, K., Guastello, F., Dunn, R., Griggs, S. A., Beasley, T. M., Gemake, J., Sinatra, R.,

& Lewthwaite, B. (1999-2000). Relationship(s) between global-format and

analytic-format learning style assessments based on the Dunn and Dunn model.

National Forum of Applied Educational Research Journal, 13(1), 76-96.


*Cambiano, R. L. (1999). Learning styles of the age cohorts: Generation X, Baby

Boomers, and the Silent Generation. (Doctoral dissertation, University of

Arkansas). Dissertation Abstracts International, 60(06), 1864A.

*Campbell, N. A. (2002). Learning style preferences of women incarcerated in a

midwestern, minimum security prison. (Doctoral dissertation, University of

Arkansas). Dissertation Abstracts International, 63(05), 1804A.

*Clark-Thayer, S. (1987). The relationship of the knowledge of student perceived

learning style preferences, and study habits and attitudes to achievement of

college freshmen in a small, urban university. (Doctoral dissertation, Boston

University). Dissertation Abstracts International, 48(04), 872A.

Cohen, J. (1977). Statistical power for the behavioral sciences. (Re. ed.). New York:

Academic Press.

Cohen, J. (1988). Statistical power for the behavioral sciences. (2nd ed.). Hillsdale, NJ:

Earlbaum.

Cohen, J. (1992). A power printer. Psychological Bulletin. 112, 155-159.

*Coolidge-Parker, J.A. (1989). A comparison of perceived and objectively measured

perceptual learning style modality elements of court reporters and court reporting

students. (Doctoral dissertation, University of South Florida). Dissertation

Abstracts International, 50(07), 1896A.

Cooper, H. & Hedges, L.V. (1994). The handbook of research synthesis. New York:

Russell Sage Foundation.

Dunn, R., Beaudry, J., & Klavas, A. (1989) Survey of research on learning styles.

Educational Leadership, 46(6), 50-58.


Dunn, R. & Dunn, K. (1972). Practical approaches to individualizing instruction. West

Nyack, New York: Parker Publishing Company Division of Prentice-Hall

Publishers, Inc.

Dunn, R. & Dunn, K. (1992). Teaching elementary students through their individual

learning styles. Boston: Allyn & Bacon.

Dunn, R. & Dunn, K. (1993). Teaching secondary students through their individual

learning styles. Boston: Allyn & Bacon.

Dunn, R. & Dunn, K. (1998). Practical approaches to individualizing staff development

for adults. Westport, CT: Praeger.

Dunn, R., & Dunn, K. (1999). The complete guide to the learning styles inservice system.

Boston: Allyn & Bacon.

Dunn, R., Dunn, K., & Price, G. (1977). Diagnosing Learning Styles: A prescription for

avoiding malpractice suits against school systems. Kappan, 418-420.

Dunn, R., Dunn, K., & Price, G. (1975, 1978, 1984, 1986, 1987, 1989, 1990, 1996).

Learning Style Inventory. Lawrence, KS: Price Systems Inc.

Dunn, R., Dunn, K., & Price, G. (1981, 1986, 1989, 1990, 1991, 1993, 1996).

Productivity Environmental Preference Survey. Lawrence, KS: Price Systems

Inc.

Dunn, R. & Griggs, S. (Eds.) (2003). Synthesis of the Dunn and Dunn Learning-Style

Model Research: Who, What, When, Where, and So What? St. John’s University.

Dunn, R., Griggs, S. A., Olson, J., Gorman, B., & Beasley, M. (1995). A meta-analytic

validation of the Dunn and Dunn learning styles model. Journal of Educational

Research, 88(6), 353-362.


Dunn, R., Thies, A. P. & Honigfeld, A. (2001). Synthesis of the Dunn and Dunn Learning

Style Model Research: Analysis from a Neuropychological Perspective. St. John’s

University.

*Eitington, N. J. (1989). A comparison of learning styles of freshmen with high and low

reading achievement in the Community Liberal Studies Program at Georgetown

University. (Doctoral dissertation, The George Washington University).

Dissertation Abstracts International, 50(05), 1285A.

Faith, M. S. Allison, D. B. & and Gorman, B. S. (1997). Meta-analysis of Single-Case

Research. In R. D. Franklin, D. B. Allison & B. S. Gorman (Eds.), Design and

Analysis of Single-Case Research (p. 245-277). Mahwah, NJ: Erlbaum.

Fine, D. (2002). Comparison between the learning style of special education and regular

education high school students and the effects of responsive teaching on the short-

and long-term achievement, attitudes, and behaviors of a subset of SPED

adolescents. (Doctoral dissertation, St. John’s University). Dissertation Abstracts


*Foley, N. E. (1997). Learning style preferences of undergraduate students with and

without learning disabilities. (Doctoral dissertation, University of Missouri-

Columbia). Dissertation Abstracts International, 59(7), 2443A.

*Franchi, J. (2002). Comparison of learning and performance styles of cross-cultural

populations in a global corporate organization (Puerto Rico, Columbia, United

States. (Doctoral dissertation, The Florida State University). Dissertation


*Galvin, A. J. (1992). An analysis of learning and productivity styles across occupational


groups in a corporate setting (learning styles, corporate training). (Doctoral

dissertation, Boston University). Dissertation Abstracts International, 53(04),

1027A.

*Given, B., Knight, G. D., Patrick, S., & McGuire, N. (1999). Personality types and

learning styles of college freshman. Educational Psychology Review.

Glass, G.V. (1976). Primary, secondary, and meta-analysis. Educational Researcher, 5,

3-8.

Glass, G., McGaw, B., & Smith, M. L. (1981) Meta-analysis in social research.

Beverly Hills, CA: Sage.

*Gordon, H. R. (1996, April). Analysis of productivity and learning-style preferences of

participants in distance education. Paper presented at the meeting of the

American Education Research Association, New York.

Gorman, B. S., Primavera, L. H. & Allison, D. B. (1995). POWPAL: A program for

estimating effect sizes, statistical power and sample sizes. Educational and

Psychological Measurement. 55, 773- 776.

*Gould, B. J. (1987). An investigation of the relationship between supervisors’ and

supervisee’ sociological productivity styles and teacher evaluations and

interpersonal attraction ratings. (Doctoral dissertation, St. John’s University).


*Griggs, S. A., Price, G. E., & Suh, B. (1997/98). Freshman's level of satisfaction with

roommates, residence, and college in relation to their learning-style preferences.

National Forum of Teacher Education Journal, Los Angeles, CA: California State

University, 8(2), 29-37.


*Harrelson, G. L., Leaver-Dunn, D., & Wright, K. E. (1998). An assessment of learning

styles among undergraduate athletic training students. Journal of Athletic

Training. 33, 50-53.

*Hawk, T. D. (1983). A comparison of teachers’ preference for specific inservice activity

approaches and their measured learning styles. (Doctoral dissertation, Kansas

State University). Dissertation Abstracts International, 44(12), 3557A.

Hedges, L. V. (1982). Estimation of effect size from a series of independent experiments.

Psychological Bulletin, 92, 490-499.

*Hickerson-Roberts, V. L. (1983). Reading achievement, reading attitudes, self-concept,

learning styles and estimated high school grade point average as predictors of

academic success for 55 adult learners at Kansas State University. (Doctoral

dissertation, Kansas State University). Dissertation Abstracts International,

44(05), 1295A.

Hunter, J. E. & Schmidt, F. L. (1990). Methods of meta-analysis: Correcting error and

bias in research findings. New Park, CA: Sage.

Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating

research findings across studies. Beverly Hills, CA: Sage.

*Jenkins, C. (1991). The relationship between selected demographic variables and

learning environmental preferences of freshman students of Alcorn State

University. (Doctoral dissertation, The University of Mississippi). Dissertation


*Jones, D. E. (1994). Predicting college grade pint average: The contribution of four

developmental domains. (Doctoral dissertation, University of Kansas).



*Katzowitz, E. C. (2002). Predominant learning styles and multiple intelligences of

postsecondary allied health students. (Doctoral dissertation, University of

Georgia). Dissertation Abstract International, 63(11), 3852A.

Keefe, J.W. (1982). Student learning styles and brain behavior. Reston, Virginia:

National Association of Secondary School Principals.

*Kizilay, P. E. (1991). The relationship of learning style preferences and perceptions of

college climate and performance on the National Council Licensure Examination

for Registered Nurses in associate degree nursing programs (nursing students).

(Doctoral dissertation, University of Georgia). Dissertation Abstracts


*Kuznar, E. Falciglia, G. A., Wood, L., Frankel, J. (1991). Learning style preferences: a

comparison of younger and older females. Journal of Nutrition for the Elderly,

10(3), 21-32.

*Lam-Phoon, S. (1986). A comparative study of the learning styles of Southeast Asian

and American Caucasian college students of Two Seventh-day Adventist

campuses. (Doctoral dissertation, Andrews University). Dissertation Abstracts


*Li, T. C. (1989). The learning styles of the Filipino graduate students of the Evangelical

seminaries in Metro Manila. (Unpublished doctoral dissertation, Asia Graduate

School of Theology, Manila, Philippines.)

Light, R. J. & Pillemer, D. B. (1984). Summing up: the science of reviewing research.

Cambridge, MA: Harvard University Press.


Lipsey, M. W. & Wilson, D. B. (2001). Practical Meta-analysis. Thousand Oaks, CA:

Sage Publications.

*Little, T. S. (1993). The relationship of right brain/left brain hemispheric dimensions of

cognitive style between teachers and principals in Northeast Tennessee. (Doctoral

dissertation, East Tennessee State University). Dissertation Abstracts


Mahoney, M. J. (1985). Open exchange and epistemic progress. American Psychologist,

40, 29-39.

Marcus, L. (1977). How teachers view student learning styles. NASSP Bulletin, 61(408),

112-114.

*McFarland, M. B. (1989). An analysis of the relationship between learning style

perceptual preferences and attitudes toward computer-assisted instruction.

(Doctoral dissertation, Portland State University). Dissertation Abstracts


*Montgomery, F. L. (1993). A comparison of learning styles of traditional high school

students and adult students in Missouri area vocational-technical schools.

(Doctoral dissertation, University of Missouri-Columbia). Dissertation Abstracts


Oberer, J. (1999). Practical application of Thies’ philosophical and theoretical

interpretation of the biological basis of learning style and brain behavior and their

effects on the academic achievement, attitude, and behavior of fourth grade

students in a suburban school district. (Doctoral dissertation, St. John’s

University). Dissertation Abstracts International, 48(02), 283A.


Oberer, J. (2003). Effects of learning-style teaching on elementary students’

behaviors, achievement, and attitudes. Academic Exchange Quarterly, Spring

2003.

*Onwuegbuzie, A. J. & Jaio, Q. G. (1998). The relationship between library anxiety and

learning styles among graduate students for library instruction. Library and

Information Science Research, 20(3), 235-249.

Price, G., Dunn, R., & Dunn, K. (1982). Productivity Environmental Preference Manual.

Lawrence, KS: Price Systems, Inc.

*Ranne, T. M. (1996). Hawthorne un-capped: The relationship of adult learning styles to

the academic achievement of nursing students. (Doctoral dissertation, State

University of New York at Buffalo). Dissertation Abstracts International, 57(09),

3771A.

Reckinger, N. (1979, January). Choice as a way to quality learning. Educational Leadership, 36(4), 255-256. Research Based on the Dunn and Dunn Model (2001, 2003). New York: St. John’s

University Center for the Study of Learning and Teaching Styles.

*Reed, B. J. (1993). Concurrent validity of a self-report instrument to a performance-

based instrument of perceptual learning style in a vocational evaluation (learning

styles). (Doctoral dissertation, University of Northern Colorado). Dissertation


*Reynolds, J. (1988). A study of the pattern of learning style characteristics for adult

dependent decision makers. (Doctoral dissertation, Virginia Polytechnic Institute

and State University). Dissertation Abstracts International, 50(04), 854A.


*Ritchey, J. P. (1994). A study of the relationship between information processing style

and productivity environmental preference. (Doctoral dissertation, Southwestern

Baptist Theological Seminary). Dissertation Abstracts International, 55(06),

1450A.

Rosenthal, R. (1979). The file drawer problem and tolerance for null results.

Psychological Bulletin, 86, 638-641.

Rosenthal, R. & Rubin, D. (1982). A simple, general purpose display of magnitude of

experimental effect. Journal of Educational Psychology, 74, 707-712.

Rundle, S. & Dunn, R. (1996-1999). Building Excellence (BE). Pittsford, NY:

Performance Concepts.

Schmidt, F. L., & Hunter, J. E. (1977). Development of a general solution to the problem

of validity generalization. Journal of Applied Psychology, 62, 529-540.

Schwarzer, R. (1989). Meta-analysis computer program. Berlin, West Germany.

Schwarzer, R. (1996). Meta-analysis programs. Unpublished manuscript, Gottingen,

Hogrefe.

*Shaver, V. E. (2000). Learning styles and student success in radiology education.

(Doctoral dissertation, Florida Atlantic University). Dissertation Abstracts

International, 61(09), 4676B.

*Siebenman, J. B. (1984). An investigation into the relationship between learning style

and cognitive style in nontraditional college reading students. (Doctoral

dissertation, Arizona State University). Dissertation Abstracts International,

45(06), 1705A.

*Skinner, M. J. (1995). The use of the Productivity Environmental Preferences Survey


and the Minnesota Satisfaction Questionnaire to determine if any relationships

exist between learning style and job satisfaction among registered and licensed

practical nurses in acute care settings. (Doctoral dissertation, The University of

Mississippi). Dissertation Abstracts International, 56(09), 4804B.

Tendy, S. M. & Geiser, W. F. (1998-99). The search for style: It all depends on where

you look. National Forum of Teacher Education Journal, 9(1), 3-15.

Thies, A. P. (1979). A brain-behavior analysis of learning styles. In Student learning

styles: Diagnosing and prescribing programs (pp. 55-61). Reston, VA: National

Association of Secondary School Principals.

Thies, A. P. (1999-2000). The neuropsychology of learning styles. National Forum of

Applied Educational Research Journal, 13(1), 50-62.

*Tiller, R. W. (1991). An investigation into the extent of congruency between general

and music-specific learning style preferences. (Doctoral dissertation, Michigan

State University). Dissertation Abstracts International, 52(05), 1677A.

*Toppins, A. D. & Dunlap, W. R. (1984). Learning Styles: Do They Affect Faculty

Evaluation? Paper presented at the annual meeting of the Midsouth Educational

Research Association (13th, New Orleans, LA, November 14-16, 1984).

*Vafa, S. G. (2002). Learning style preferences among University of Houston online

students (Texas) (Doctoral dissertation, University of Houston). Dissertation


*Van Wynen (1999). The relationships between selected learning-style elements and the

achievement and attitudes of older adults following wellness sessions using

different instructional methodologies. (Doctoral dissertation, St. John’s


University). Dissertation Abstract International, 62(05), 1740A.

*Vazquez, A. W. (1985). Description of learning styles of high-risk adult students taking

evening courses in urban community colleges in Puerto Rico. (Doctoral

dissertation, The Union for Experimenting Colleges and Universities).


*Wallin, P. D. K. (1998). Teaching and learning style factors of National Board-certified

teachers. (Doctoral dissertation, The University of Southern Mississippi).


*Wegner, W. W. (1980). Opsimathic styles of adults. (Doctoral dissertation, The

University of Southern Mississippi). Dissertation Abstracts International, 41(05),

1898A.

*White, R. H. (1993). A study of the relationships between teachers’ learning style

preferences and their use of the microcomputer at home and at school in

Medfield, Massachusetts. (Doctoral dissertation, Boston College). Dissertation


*Williams, E. H. Jr. (1986). The learning style characteristics of typical and

developmental junior college students. (Doctoral dissertation, The University of

Southern Mississippi). Dissertation Abstracts International, 47(11), 4036A.

*Wittenberg, S. K. (1984). A comparison of diagnosed and preferred learning styles of

young adults in need of remediation. (Doctoral dissertation, The University of

Toledo). Dissertation Abstracts International, 45(12), 3539A.

Wolf, F. (1986). Meta-analysis: Quantitative methods for research synthesis. Beverly

Hills, CA: Sage Publications.


*Wooldridge, B. E. M. (1993). A strategy for improving management education for

public administrators: Integrating the results of selected learning style research

into public administration course design and delivery. (Doctoral dissertation,

University of Southern California). Dissertation Abstracts International, 55(12),

3988A.

* Research included in this meta-analysis


Figure 1. Dunn and Dunn Learning-Style Model


Table 1.

Meta-analysis Results for Mean Effect Sizes

Simple

Unweighted

Schmidt-Hunter

Fisher’s Z

Transformation

Fisher's Z

Schmidt-Hunter

Unweighted r

.258

.269

Weighted r

.236

.241

95% Confidence Interval Low

.247

.082

.256

.234

95%

Confidence

Interval

High

.269

.391

.270

.247


Table 2.

Mean Effect Sizes for Variables Examined

VARIABLE (N)

MEAN EFFECT SIZE

95% CONFIDENCE INTERVAL-LOW

95% CONFIDENCE INTERVAL-HIGH

Discipline (13)

.363

.290

.435

Achievement (22)

.351

.285

.418

Decision-Making Group (4)

.343

.290

.396

Age (49)

.326

.291

.360

Ethnicity (12)

.311

.243

.379

Right Brain/Left Brain (16)

.303

.256

.319

Gender (42)

.288

.256

.319

Disability (7)

.277

.259

.296

Faculty Evaluation (9)

.269

.116

.423

Study Habits (15)

.252

.117

.331

School (6)

.247

.185

.309

Employment Status (7)

.247

.230

.264


Judging/Perceiving (13)

.241

.199

.282

Educational Attainment (10)

.240

.162

.318

Type of Program (25)

.236

.187

.285

Field Dependence/Independence (11)

.236

.210

.261

Computer-Use (7)

.216

.169

.263

Job/School Satisfaction (5)

.214

.177

.251

Sensing/Intuition (8)

.212

.171

.254

Academic Level (18)

.204

.140

.268

GPA (20)

.191

.158

.224

Profession (32)

.190

.112

.443

Study Attitude (9)

.184

.124

.245

Parent-Status (5)

.182

.144

.220


Library Anxiety (6)

.170

.150

.190

Entrance Exam (1)

.166

N/A

N/A

Extrovert/Introvert (6)

.165

.109

.221

Thinking/Feeling (7)

.146

.125

.166

SES (1)

.126

N/A

N/A

Note. N = Total number of effect sizes for each variable. N/A = 95% confidence interval cannot be calculated for one effect size.


Table 3.

Achievement Correlations

ELEMENT

HIGH ACHIEVERS

LOW ACHIEVERS

AUDITORY

.395 (1)

KINESTHETIC

.637 (2)

NOT VISUAL

.419 (2)

TACTUAL

.339 (2)

.228 (1)

RESPONSIBILITY

.190 (1)

MOTIVATION

.249 (2)

STRUCTURE

.382 (1)

AUTHORITY

.281 (2)

PEERS

.250 (1)


SEVERAL WAYS

.312 (2)

SOUND

.400 (1)

LIGHT

.450 (1)

DESIGN

.470 (1)

INTAKE

.150 (1)

Note: (N) = number of individual effect sizes


Table 4.

Age Correlations

ELEMENT

UNDER 30

OVER 30

OVER 55

KINESTHETIC

.210

TEMPERATURE

.197

.354

LEARN ALONE

.251

PEERS

.310

.379

STRUCTURE

.209

.361

.785

INTAKE

.265

.279

DESIGN

.337

.286

SELF-

MOTIVATED

.361

.326


PARENT-

MOTIVATED

.180

TEACHER-

MOTIVATED

.315

MORNING

.343

AFTERNOON

.362

LATE MORNING

.267

VISUAL

.482

.270

LIGHT

.308

.343

AUTHORITY

.289

.574

RESPONSIBILITY

.347

.491

TACTUAL

.228

.374


SOUND

.408

SEVERAL WAYS

.538

MOBILITY

.427

Note: All mean effect sizes were based on one effect size.


Table 5.

Attitude Correlations

ELEMENT

SATISFIED

NOT

SATISFIED

POSITIVE

NEGATIVE

PERSISTENT

.200 (1)

RESPONSIBILTY

.229 (1)

.194 (1)

MOTIVATED

.260 (1)

.327 (1)

LEARN ALONE

.194 (1)

.149 (1)

QUIET

.189 (1)

COMBINATION

OF ELEMENTS

.197 (1) (quiet,

cooler

temperatures,

bright light,

less motivated,

more

responsible)


LIGHT

.106 (1)

DESIGN

.123 (1)

REFLECTIVE

.178 (1)

VISUAL

.176 (1)

INTAKE

.109 (1)

MOBILITY

.296(1)



Table 6.

Computer Usage Correlations for Individuals with High Computer Usage

ELEMENT

MEAN EFFECT SIZE

KINESTHETIC (2)

.238

VISUAL (2)

.222

TACTUAL (1)

.232

TEMPERATURE (1)

.186

MORNING (1)

.291

ALL PERCEPTUAL

STRENGTHS (1)

.148


Table 7.

Discipline Correlations

ELEMENT

TEACHERS

NURSES

MUSIC

STUDENTS

BUSINESS

STUDENTS

AUDITORY

.524 (1)

KINESTHETIC

.310 (1)

.388 (1)

.612 (1)

VISUAL

.332 (1)

.420 (1)

TACTUAL

.313 (1)

AUTHORITY

.220 (1)

.334 (1)

MOBILITY

.313 (1)

.293(1)

RESPONSIBILITY

.336 (1)

PEERS

.235 (2)

.277 (1)

MORNING

.386 (1)

.315 (1)


INTAKE

.279 (1)

.549 (1)

LIGHT

.379 (2)

TEMPERATURE

.197 (1)

SOUND

.400 (1)

AFTERNOON

.362 (1)

MOTIVATED

.299 (2)

REFLECTIVE

.338 (1)

ALL ELEMENTS

.206 (1)



Table 8.

Ethnicity Correlations

Element

Caucasian

Non-

Caucasian

Caucasian

Compared

with

Asian

Caucasian

Compared

with African-

American

African

American

Hispanic

DESIGN

.358 (1)

MORNING

.258 (1)

AFTERNOON

.243 (1)

AUDITORY

.318 (1)

.303 (1)

VISUAL

.287 (1)

KINESTHETIC

.325 (1)

SOUND

.226 (1)

INTAKE

.251 (2)


ALL

ELEMENTS

.279 (1)

.627 (1)



Table 9.

Gender Correlations

ELEMENT

FEMALES

MALES

NOT

SPECIFIED

KINESTHETIC

.214 (2)

VISUAL

.238 (2)

AUDITORY

.225 (3)

TACTUAL

.362 (2)

PERSISITENCE

.169 (1)

.353 (2)

RESPONSIBILITY

.275 (1)

.273 (1)

MOTIVATION

.202 (2)

STRUCTURE

.142 (1)

AUTHORITY

.275 (5)


LEARN ALONE

.198 (1)

.287 (1)

PEERS

.276 (2)

SEVERAL WAYS

.277 (1)

.197 (1)

TEMPERATURE

.389 (1)

DESIGN

.343 (2)

.474 (1)

MORNING

.249 (1)

.174 (1)

AFTERNOON

.374 (1)

.179 (1)

INTAKE

.440 (3)

MOBILITY

.353 (2)

.297 (1)

ALL ELEMENTS

.358 (1)



Table 10.

GPA Correlations

ELEMENT

HIGH

GPA

LOW

GPA

HIGH COMBINED

ACHIEVEMENT

LOW

COMBINED

ACHIEVEMENT

KINESTHETIC

.143 (1)

.472 (3)

AUDITORY

.395 (1)

VISUAL

.419 (2)

NOT VISUAL

.224 (1)

TACTUAL

.339 (2)

.228 (1)

NOT TACTUAL

.224 (1)

RESPONSIBILITY

.226 (3)

.217 (4)

MOTIVATION

.220 (2)

.324 (4)


STRUCTURE

.224 (2)

.276 (3)

NO AUTHORITY-

FIGURE PRESENT

.164 (1)

.164 (1)

AUTHORITY

.281 (2)

LEARN ALONE

.280 (2)

.280 (2)

PEERS

.250 (1)

AFTERNOON

.109 (1)

.109 (1)

SEVERAL WAYS

.164 (1)

.263 (3)

INTAKE

.134 (1)

.143 (2)

SOUND

.127 (1)

.400 (1)

.127 (1)

DESIGN

.166 (1)



Table 11.

Profession Correlations

ELEMENT

EDUCATORS

CORPORATE

EMPLOYEES

NURSES

PARALEGALS

AUDITORY

.113 (1)

.191 (1)

KINESTHETIC

.195 (1)

.235 (1)

VISUAL

.133 (2)

.232 (1)

TACTUAL

.163 (1)

(N0T

TACTUAL)

.168 (1)

SEVERAL WAYS

.187 (1)

.234 (1)

SOUND

.198 (1)

LIGHT

.226 (1)

DESIGN

.116 (1)

.188 (1)


INTAKE

.170 (1)

.246 (1)

MOBILITY

.217 (2)

PERSISTENCE

.154 (1)

RESPONSIBILITY

.154 (1)

MOTIVATED

.118 (1)

AUTHORITY

.201 (2)

.195 (1)

LEARN ALONE

.176 (1)

PEERS

.279 (1)

AFTERNOON

.146 (1)

.183 (1)

NOT LATE

MORNING

.149 (1)


introduction - ericgriggs, olson, beasley, & gorman, 1995). the dunn and dunn learning-style...

Documents